Automatic Speech Recognition
NeMo
PyTorch
English
speech
audio
Transducer
FastConformer
Conformer
NeMo
hf-asr-leaderboard
Eval Results

Context Size of RNNT variants of Parakeet

#1
by puvvadasaikiran - opened

any particular bounds on context size of Parakeet like their RNNT counterparts (Meta's Emformer RNNT )

NVIDIA org

Parakeet uses Fast Conformer encoder. In limited context mode, we've done inference on upto 8 hours audio sample (synthetically concatenated) in a single forward pass on a single 80GB A100

Sign up or log in to comment